Skip to main content

cargo/sources/git/
utils.rs

1//! Utilities for handling git repositories, mainly around
2//! authentication/cloning.
3
4use crate::context::ProgressWhen;
5use crate::sources::git::fetch::RemoteKind;
6use crate::sources::git::oxide;
7use crate::sources::git::oxide::cargo_config_to_gitoxide_overrides;
8use crate::sources::git::source::GitSource;
9use crate::sources::source::Source as _;
10use crate::util::HumanBytes;
11use crate::util::errors::{CargoResult, GitCliError};
12use crate::util::network::http::http_handle;
13use crate::util::network::http::needs_custom_http_transport;
14use crate::util::{GlobalContext, IntoUrl, MetricsCounter, Progress, network};
15use crate::workspace::{GitReference, SourceId};
16
17use anyhow::{Context as _, anyhow};
18use cargo_util::{ProcessBuilder, Stdio, paths};
19use cargo_util_terminal::Verbosity;
20use git2::{ErrorClass, ObjectType, Oid};
21use http::{Request, StatusCode};
22use tracing::{debug, info};
23use url::Url;
24
25use std::borrow::Cow;
26use std::path::{Path, PathBuf};
27use std::process::Command;
28use std::str;
29use std::sync::Once;
30use std::sync::atomic::{AtomicBool, Ordering};
31use std::time::{Duration, Instant};
32
33/// A file indicates that if present, `git reset` has been done and a repo
34/// checkout is ready to go. See [`GitCheckout::reset`] for why we need this.
35const CHECKOUT_READY_LOCK: &str = ".cargo-ok";
36
37/// A remote repository. It gets cloned into a local [`GitDatabase`].
38#[derive(PartialEq, Clone, Debug)]
39pub struct GitRemote {
40    /// URL to a remote repository.
41    ///
42    /// This may differ from the [`SourceId`] URL when the original URL
43    /// can't be represented as a WHATWG [`Url`], for example SCP-like URLs.
44    /// See <https://github.com/rust-lang/cargo/issues/16740>.
45    url: String,
46}
47
48/// A local clone of a remote repository's database. Multiple [`GitCheckout`]s
49/// can be cloned from a single [`GitDatabase`].
50pub struct GitDatabase {
51    /// The remote repository where this database is fetched from.
52    remote: GitRemote,
53    /// Path to the root of the underlying Git repository on the local filesystem.
54    path: PathBuf,
55    /// Underlying Git repository instance for this database.
56    repo: git2::Repository,
57}
58
59/// A local checkout of a particular revision from a [`GitDatabase`].
60pub struct GitCheckout<'a> {
61    /// The git database where this checkout is cloned from.
62    database: &'a GitDatabase,
63    /// Path to the root of the underlying Git repository on the local filesystem.
64    path: PathBuf,
65    /// The git revision this checkout is for.
66    revision: git2::Oid,
67    /// Underlying Git repository instance for this checkout.
68    repo: git2::Repository,
69}
70
71impl GitRemote {
72    /// Creates an instance for a remote repository URL.
73    pub fn new(url: &Url) -> GitRemote {
74        GitRemote {
75            url: url.as_str().to_owned(),
76        }
77    }
78
79    /// Creates an instance with an URL that may not be a valid WHATWG URL.
80    ///
81    /// This is needed because [`SourceId`] hasn't yet supported SCP-like URLs.
82    pub(super) fn new_from_str(url: String) -> GitRemote {
83        GitRemote { url }
84    }
85
86    /// Gets the remote repository URL.
87    pub fn url(&self) -> &str {
88        &self.url
89    }
90
91    /// Fetches and checkouts to a reference or a revision from this remote
92    /// into a local path.
93    ///
94    /// This ensures that it gets the up-to-date commit when a named reference
95    /// is given (tag, branch, refs/*). Thus, network connection is involved.
96    ///
97    /// If we have a previous instance of [`GitDatabase`] then fetch into that
98    /// if we can. If that can successfully load our revision then we've
99    /// populated the database with the latest version of `reference`, so
100    /// return that database and the rev we resolve to.
101    pub fn checkout(
102        &self,
103        into: &Path,
104        db: Option<GitDatabase>,
105        manifest_reference: &GitReference,
106        reference: &GitReference,
107        gctx: &GlobalContext,
108    ) -> CargoResult<(GitDatabase, git2::Oid)> {
109        if let Some(mut db) = db {
110            fetch(
111                &mut db.repo,
112                self.url(),
113                manifest_reference,
114                reference,
115                gctx,
116                RemoteKind::GitDependency,
117            )
118            .with_context(|| format!("failed to fetch into: {}", into.display()))?;
119
120            if let Some(rev) = resolve_ref(reference, &db.repo).ok() {
121                return Ok((db, rev));
122            }
123        }
124
125        // Otherwise start from scratch to handle corrupt git repositories.
126        // After our fetch (which is interpreted as a clone now) we do the same
127        // resolution to figure out what we cloned.
128        if into.exists() {
129            paths::remove_dir_all(into)?;
130        }
131        paths::create_dir_all(into)?;
132        let mut repo = init(into, true)?;
133        fetch(
134            &mut repo,
135            self.url(),
136            manifest_reference,
137            reference,
138            gctx,
139            RemoteKind::GitDependency,
140        )
141        .with_context(|| format!("failed to clone into: {}", into.display()))?;
142        let rev = resolve_ref(reference, &repo)?;
143
144        Ok((
145            GitDatabase {
146                remote: self.clone(),
147                path: into.to_path_buf(),
148                repo,
149            },
150            rev,
151        ))
152    }
153
154    /// Creates a [`GitDatabase`] of this remote at `db_path`.
155    pub fn db_at(&self, db_path: &Path) -> CargoResult<GitDatabase> {
156        let repo = git2::Repository::open(db_path)?;
157        Ok(GitDatabase {
158            remote: self.clone(),
159            path: db_path.to_path_buf(),
160            repo,
161        })
162    }
163}
164
165impl GitDatabase {
166    /// Checkouts to a revision at `dest`ination from this database.
167    #[tracing::instrument(skip(self, gctx))]
168    pub fn copy_to(
169        &self,
170        rev: git2::Oid,
171        dest: &Path,
172        gctx: &GlobalContext,
173        quiet: bool,
174    ) -> CargoResult<GitCheckout<'_>> {
175        // If the existing checkout exists, and it is fresh, use it.
176        // A non-fresh checkout can happen if the checkout operation was
177        // interrupted. In that case, the checkout gets deleted and a new
178        // clone is created.
179        let checkout = match git2::Repository::open(dest)
180            .ok()
181            .map(|repo| GitCheckout::new(self, rev, repo))
182            .filter(|co| co.is_fresh())
183        {
184            Some(co) => co,
185            None => {
186                let (checkout, guard) = GitCheckout::clone_into(dest, self, rev, gctx)?;
187                checkout.update_submodules(gctx, quiet)?;
188                guard.mark_ok()?;
189                checkout
190            }
191        };
192
193        Ok(checkout)
194    }
195
196    /// Get a short OID for a `revision`, 7 chars or more if ambiguous.
197    ///
198    /// Like [`git2::Object::short_id`]
199    /// but ignores the user's `core.abbrev` git config.
200    pub fn to_short_id(&self, rev: git2::Oid) -> CargoResult<String> {
201        const MIN_ABBREV_LEN: usize = 7; // this is git/libgit2's default
202        let odb = self.repo.odb()?;
203        let mut len = MIN_ABBREV_LEN;
204        let mut hex = rev.to_string();
205        // quasi- re-implementation of
206        // https://github.com/libgit2/libgit2/blob/26055f5af74ab/src/libgit2/object.c#L523-L573
207        while len < hex.len() {
208            match odb.exists_prefix(rev, len) {
209                Ok(_) => break,
210                Err(err) if err.code() == git2::ErrorCode::Ambiguous => len += 1,
211                Err(err) => return Err(err.into()),
212            }
213        }
214        hex.truncate(len);
215        Ok(hex)
216    }
217
218    /// Checks if the database contains the object of this `oid`..
219    pub fn contains(&self, oid: git2::Oid) -> bool {
220        self.repo.revparse_single(&oid.to_string()).is_ok()
221    }
222
223    /// [`resolve_ref`]s this reference with this database.
224    pub fn resolve(&self, r: &GitReference) -> CargoResult<git2::Oid> {
225        resolve_ref(r, &self.repo)
226    }
227}
228
229/// Resolves [`GitReference`] to an object ID with objects the `repo` currently has.
230pub fn resolve_ref(gitref: &GitReference, repo: &git2::Repository) -> CargoResult<git2::Oid> {
231    let id = match gitref {
232        // Note that we resolve the named tag here in sync with where it's
233        // fetched into via `fetch` below.
234        GitReference::Tag(s) => (|| -> CargoResult<git2::Oid> {
235            let refname = format!("refs/remotes/origin/tags/{}", s);
236            let id = repo.refname_to_id(&refname)?;
237            let obj = repo.find_object(id, None)?;
238            let obj = obj.peel(ObjectType::Commit)?;
239            Ok(obj.id())
240        })()
241        .with_context(|| format!("failed to find tag `{}`", s))?,
242
243        // Resolve the remote name since that's all we're configuring in
244        // `fetch` below.
245        GitReference::Branch(s) => {
246            let name = format!("origin/{}", s);
247            let b = repo
248                .find_branch(&name, git2::BranchType::Remote)
249                .with_context(|| format!("failed to find branch `{}`", s))?;
250            b.get()
251                .target()
252                .ok_or_else(|| anyhow::format_err!("branch `{}` did not have a target", s))?
253        }
254
255        // We'll be using the HEAD commit
256        GitReference::DefaultBranch => {
257            let head_id = repo.refname_to_id("refs/remotes/origin/HEAD")?;
258            let head = repo.find_object(head_id, None)?;
259            head.peel(ObjectType::Commit)?.id()
260        }
261
262        GitReference::Rev(s) => {
263            let obj = repo.revparse_single(s)?;
264            match obj.as_tag() {
265                Some(tag) => tag.target_id(),
266                None => obj.id(),
267            }
268        }
269    };
270    Ok(id)
271}
272
273impl<'a> GitCheckout<'a> {
274    /// Creates an instance of [`GitCheckout`]. This doesn't imply the checkout
275    /// is done. Use [`GitCheckout::is_fresh`] to check.
276    ///
277    /// * The `database` is where this checkout is from.
278    /// * The `repo` will be the checked out Git repository.
279    fn new(
280        database: &'a GitDatabase,
281        revision: git2::Oid,
282        repo: git2::Repository,
283    ) -> GitCheckout<'a> {
284        let path = repo.workdir().unwrap_or_else(|| repo.path());
285        GitCheckout {
286            path: path.to_path_buf(),
287            database,
288            revision,
289            repo,
290        }
291    }
292
293    /// Gets the remote repository URL.
294    fn remote_url(&self) -> &str {
295        self.database.remote.url()
296    }
297
298    /// Clone a repo for a `revision` into a local path from a `database`.
299    /// This is a filesystem-to-filesystem clone.
300    fn clone_into(
301        into: &Path,
302        database: &'a GitDatabase,
303        revision: git2::Oid,
304        gctx: &GlobalContext,
305    ) -> CargoResult<(GitCheckout<'a>, CheckoutGuard)> {
306        let dirname = into.parent().unwrap();
307        paths::create_dir_all(&dirname)?;
308        if into.exists() {
309            paths::remove_dir_all(into)?;
310        }
311
312        // we're doing a local filesystem-to-filesystem clone so there should
313        // be no need to respect global configuration options, so pass in
314        // an empty instance of `git2::Config` below.
315        let git_config = git2::Config::new()?;
316
317        // Clone the repository, but make sure we use the "local" option in
318        // libgit2 which will attempt to use hardlinks to set up the database.
319        // This should speed up the clone operation quite a bit if it works.
320        //
321        // Note that we still use the same fetch options because while we don't
322        // need authentication information we may want progress bars and such.
323        let url = database.path.into_url()?;
324        let mut repo = None;
325        with_fetch_options(&git_config, url.as_str(), gctx, &mut |fopts| {
326            let mut checkout = git2::build::CheckoutBuilder::new();
327            checkout.dry_run(); // we'll do this below during a `reset`
328
329            let r = git2::build::RepoBuilder::new()
330                // use hard links and/or copy the database, we're doing a
331                // filesystem clone so this'll speed things up quite a bit.
332                .clone_local(git2::build::CloneLocal::Local)
333                .with_checkout(checkout)
334                .fetch_options(fopts)
335                .clone(url.as_str(), into)?;
336            // `git2` doesn't seem to handle shallow repos correctly when doing
337            // a local clone. Fortunately all that's needed is the copy of the
338            // one file that defines the shallow boundary, the commits which
339            // have their parents omitted as part of the shallow clone.
340            //
341            // TODO(git2): remove this when git2 supports shallow clone correctly
342            if database.repo.is_shallow() {
343                std::fs::copy(
344                    database.repo.path().join("shallow"),
345                    r.path().join("shallow"),
346                )?;
347            }
348            repo = Some(r);
349            Ok(())
350        })?;
351        let repo = repo.unwrap();
352
353        let checkout = GitCheckout::new(database, revision, repo);
354        let guard = checkout.reset(gctx)?;
355        Ok((checkout, guard))
356    }
357
358    /// Checks if the `HEAD` of this checkout points to the expected revision.
359    fn is_fresh(&self) -> bool {
360        match self.repo.revparse_single("HEAD") {
361            Ok(ref head) if head.id() == self.revision => {
362                // See comments in reset() for why we check this
363                self.path.join(CHECKOUT_READY_LOCK).exists()
364            }
365            _ => false,
366        }
367    }
368
369    /// Similar to [`reset()`]. This roughly performs `git reset --hard` to the
370    /// revision of this checkout, with additional interrupt protection by a
371    /// dummy file [`CHECKOUT_READY_LOCK`].
372    ///
373    /// If we're interrupted while performing a `git reset` (e.g., we die
374    /// because of a signal) Cargo needs to be sure to try to check out this
375    /// repo again on the next go-round.
376    ///
377    /// To enable this we have a dummy file in our checkout, [`.cargo-ok`],
378    /// which if present means that the repo has been successfully reset and is
379    /// ready to go. Hence if we start to do a reset, we make sure this file
380    /// *doesn't* exist. The caller of [`reset`] has an option to perform additional operations
381    /// (e.g. submodule update) before marking the check-out as ready.
382    ///
383    /// [`.cargo-ok`]: CHECKOUT_READY_LOCK
384    fn reset(&self, gctx: &GlobalContext) -> CargoResult<CheckoutGuard> {
385        let guard = CheckoutGuard::guard(&self.path);
386        info!("reset {} to {}", self.repo.path().display(), self.revision);
387
388        // Ensure libgit2 won't mess with newlines when we vendor.
389        if let Ok(mut git_config) = self.repo.config() {
390            git_config.set_bool("core.autocrlf", false)?;
391        }
392
393        let object = self.repo.find_object(self.revision, None)?;
394        reset(&self.repo, &object, gctx)?;
395
396        Ok(guard)
397    }
398
399    /// Like `git submodule update --recursive` but for this git checkout.
400    ///
401    /// This function respects `submodule.<name>.update = none`[^1] git config.
402    /// Submodules set to `none` won't be fetched.
403    ///
404    /// [^1]: <https://git-scm.com/docs/git-submodule#Documentation/git-submodule.txt-none>
405    fn update_submodules(&self, gctx: &GlobalContext, quiet: bool) -> CargoResult<()> {
406        return update_submodules(&self.repo, gctx, quiet, self.remote_url());
407
408        /// Recursive helper for [`GitCheckout::update_submodules`].
409        fn update_submodules(
410            repo: &git2::Repository,
411            gctx: &GlobalContext,
412            quiet: bool,
413            parent_remote_url: &str,
414        ) -> CargoResult<()> {
415            debug!("update submodules for: {:?}", repo.workdir().unwrap());
416
417            for mut child in repo.submodules()? {
418                update_submodule(repo, &mut child, gctx, quiet, parent_remote_url).with_context(
419                    || {
420                        format!(
421                            "failed to update submodule `{}`",
422                            child.name().unwrap_or("")
423                        )
424                    },
425                )?;
426            }
427            Ok(())
428        }
429
430        /// Update a single Git submodule, and recurse into its submodules.
431        fn update_submodule(
432            parent: &git2::Repository,
433            child: &mut git2::Submodule<'_>,
434            gctx: &GlobalContext,
435            quiet: bool,
436            parent_remote_url: &str,
437        ) -> CargoResult<()> {
438            child.init(false)?;
439
440            let child_url_str = child
441                .url()
442                .with_context(|| {
443                    format!("failed to update submodule `{}`", child.path().display())
444                })?
445                .ok_or_else(|| {
446                    anyhow::format_err!(
447                        "unable to update submodule `{}` without a path",
448                        child.name().unwrap_or("")
449                    )
450                })?;
451
452            // Skip the submodule if the config says not to update it.
453            if child.update_strategy() == git2::SubmoduleUpdate::None {
454                gctx.shell().status(
455                    "Skipping",
456                    format!(
457                        "git submodule `{}` due to update strategy in .gitmodules",
458                        child_url_str
459                    ),
460                )?;
461                return Ok(());
462            }
463
464            let child_remote_url = absolute_submodule_url(parent_remote_url, child_url_str)?;
465
466            // A submodule which is listed in .gitmodules but not actually
467            // checked out will not have a head id, so we should ignore it.
468            let Some(head) = child.head_id() else {
469                return Ok(());
470            };
471
472            // If the submodule hasn't been checked out yet, we need to
473            // clone it. If it has been checked out and the head is the same
474            // as the submodule's head, then we can skip an update and keep
475            // recursing.
476            let head_and_repo = child.open().and_then(|repo| {
477                let target = repo.head()?.target();
478                Ok((target, repo))
479            });
480            let repo = match head_and_repo {
481                Ok((head, repo)) => {
482                    if child.head_id() == head {
483                        return update_submodules(&repo, gctx, quiet, &child_remote_url);
484                    }
485                    repo
486                }
487                Err(..) => {
488                    let path = parent.workdir().unwrap().join(child.path());
489                    let _ = paths::remove_dir_all(&path);
490                    init(&path, false)?
491                }
492            };
493            // Fetch submodule database and checkout to target revision
494            let reference = GitReference::Rev(head.to_string());
495
496            // SCP-like URL is not a WHATWG Standard URL.
497            // `url` crate can't parse SCP-like URLs.
498            // We convert to `ssh://` for SourceId,
499            // but preserve the original URL for fetch to maintain correct semantics
500            // See <https://github.com/rust-lang/cargo/issues/16740>
501            let (source_url, fetch_url) = match child_remote_url.as_ref().into_url() {
502                Ok(url) => (url, None),
503                Err(_) => {
504                    let ssh_url = scp_to_ssh(&child_remote_url)
505                        .ok_or_else(|| anyhow::format_err!("invalid url `{child_remote_url}`"))?
506                        .as_str()
507                        .into_url()?;
508                    (ssh_url, Some(child_remote_url.into_owned()))
509                }
510            };
511
512            // GitSource created from SourceId without git precise will result to
513            // locked_rev being Deferred and fetch_db always try to fetch if online
514            let source_id =
515                SourceId::for_git(&source_url, reference)?.with_git_precise(Some(head.to_string()));
516
517            let mut source = match &fetch_url {
518                Some(url) => GitSource::new_for_submodule(source_id, url.to_owned(), gctx)?,
519                None => GitSource::new(source_id, gctx)?,
520            };
521            source.set_quiet(quiet);
522
523            let (db, actual_rev) = source.fetch_db(true).with_context(|| {
524                let name = child.name().unwrap_or("");
525                let url = fetch_url.unwrap_or_else(|| source_url.to_string());
526                format!("failed to fetch submodule `{name}` from {url}")
527            })?;
528            db.copy_to(actual_rev, repo.path(), gctx, quiet)?;
529            Ok(())
530        }
531    }
532}
533
534/// See [`GitCheckout::reset`] for rationale on this type.
535#[must_use]
536struct CheckoutGuard {
537    ok_file: PathBuf,
538}
539
540impl CheckoutGuard {
541    fn guard(path: &Path) -> Self {
542        let ok_file = path.join(CHECKOUT_READY_LOCK);
543        let _ = paths::remove_file(&ok_file);
544        Self { ok_file }
545    }
546
547    fn mark_ok(self) -> CargoResult<()> {
548        let _ = paths::create(self.ok_file)?;
549        Ok(())
550    }
551}
552
553/// Constructs an absolute URL for a child submodule URL with its parent base URL.
554///
555/// Git only assumes a submodule URL is a relative path if it starts with `./`
556/// or `../` [^1]. To fetch the correct repo, we need to construct an absolute
557/// submodule URL.
558///
559/// At this moment it comes with some limitations:
560///
561/// * GitHub doesn't accept non-normalized URLs with relative paths.
562///   (`ssh://git@github.com/rust-lang/cargo.git/relative/..` is invalid)
563/// * `url` crate cannot parse SCP-like URLs.
564///   (`git@github.com:rust-lang/cargo.git` is not a valid WHATWG URL)
565///
566/// To overcome these, this patch always tries [`Url::parse`] first to normalize
567/// the path. If it couldn't, append the relative path and/or convert SCP-like URLs
568/// to ssh:// format as the last resorts and pray the remote git service supports
569/// non-normalized URLs.
570///
571/// See also rust-lang/cargo#12404 and rust-lang/cargo#12295.
572///
573/// [^1]: <https://git-scm.com/docs/git-submodule>
574fn absolute_submodule_url<'s>(base_url: &str, submodule_url: &'s str) -> CargoResult<Cow<'s, str>> {
575    let absolute_url = if ["./", "../"].iter().any(|p| submodule_url.starts_with(p)) {
576        match Url::parse(base_url) {
577            Ok(mut base_url) => {
578                let path = base_url.path();
579                if !path.ends_with('/') {
580                    base_url.set_path(&format!("{path}/"));
581                }
582                let absolute_url = base_url.join(submodule_url).with_context(|| {
583                    format!(
584                        "failed to parse relative child submodule url `{submodule_url}` \
585                        using parent base url `{base_url}`"
586                    )
587                })?;
588                Cow::from(absolute_url.to_string())
589            }
590            Err(_) => {
591                let mut absolute_url = base_url.to_string();
592                if !absolute_url.ends_with('/') {
593                    absolute_url.push('/');
594                }
595                absolute_url.push_str(submodule_url);
596                Cow::from(absolute_url)
597            }
598        }
599    } else {
600        Cow::from(submodule_url)
601    };
602
603    Ok(absolute_url)
604}
605
606/// Converts an SCP-like URL to `ssh://` format.
607fn scp_to_ssh(url: &str) -> Option<String> {
608    let mut gix_url = gix::url::parse(gix::bstr::BStr::new(url.as_bytes())).ok()?;
609    if gix_url.serialize_alternative_form && gix_url.scheme == gix::url::Scheme::Ssh {
610        gix_url.serialize_alternative_form = false;
611        Some(gix_url.to_bstring().to_string())
612    } else {
613        None
614    }
615}
616
617/// Prepare the authentication callbacks for cloning a git repository.
618///
619/// The main purpose of this function is to construct the "authentication
620/// callback" which is used to clone a repository. This callback will attempt to
621/// find the right authentication on the system (without user input) and will
622/// guide libgit2 in doing so.
623///
624/// The callback is provided `allowed` types of credentials, and we try to do as
625/// much as possible based on that:
626///
627/// * Prioritize SSH keys from the local ssh agent as they're likely the most
628///   reliable. The username here is prioritized from the credential
629///   callback, then from whatever is configured in git itself, and finally
630///   we fall back to the generic user of `git`.
631///
632/// * If a username/password is allowed, then we fallback to git2-rs's
633///   implementation of the credential helper. This is what is configured
634///   with `credential.helper` in git, and is the interface for the macOS
635///   keychain, for example.
636///
637/// * After the above two have failed, we just kinda grapple attempting to
638///   return *something*.
639///
640/// If any form of authentication fails, libgit2 will repeatedly ask us for
641/// credentials until we give it a reason to not do so. To ensure we don't
642/// just sit here looping forever we keep track of authentications we've
643/// attempted and we don't try the same ones again.
644fn with_authentication<T, F>(
645    gctx: &GlobalContext,
646    url: &str,
647    cfg: &git2::Config,
648    mut f: F,
649) -> CargoResult<T>
650where
651    F: FnMut(&mut git2::Credentials<'_>) -> CargoResult<T>,
652{
653    let mut cred_helper = git2::CredentialHelper::new(url);
654    cred_helper.config(cfg);
655
656    let mut ssh_username_requested = false;
657    let mut cred_helper_bad = None;
658    let mut ssh_agent_attempts = Vec::new();
659    let mut any_attempts = false;
660    let mut tried_sshkey = false;
661    let mut url_attempt = None;
662
663    let orig_url = url;
664    let mut res = f(&mut |url, username, allowed| {
665        any_attempts = true;
666        if url != orig_url {
667            url_attempt = Some(url.to_string());
668        }
669        // libgit2's "USERNAME" authentication actually means that it's just
670        // asking us for a username to keep going. This is currently only really
671        // used for SSH authentication and isn't really an authentication type.
672        // The logic currently looks like:
673        //
674        //      let user = ...;
675        //      if (user.is_null())
676        //          user = callback(USERNAME, null, ...);
677        //
678        //      callback(SSH_KEY, user, ...)
679        //
680        // So if we're being called here then we know that (a) we're using ssh
681        // authentication and (b) no username was specified in the URL that
682        // we're trying to clone. We need to guess an appropriate username here,
683        // but that may involve a few attempts. Unfortunately we can't switch
684        // usernames during one authentication session with libgit2, so to
685        // handle this we bail out of this authentication session after setting
686        // the flag `ssh_username_requested`, and then we handle this below.
687        if allowed.contains(git2::CredentialType::USERNAME) {
688            debug_assert!(username.is_none());
689            ssh_username_requested = true;
690            return Err(git2::Error::from_str("gonna try usernames later"));
691        }
692
693        // An "SSH_KEY" authentication indicates that we need some sort of SSH
694        // authentication. This can currently either come from the ssh-agent
695        // process or from a raw in-memory SSH key. Cargo only supports using
696        // ssh-agent currently.
697        //
698        // If we get called with this then the only way that should be possible
699        // is if a username is specified in the URL itself (e.g., `username` is
700        // Some), hence the unwrap() here. We try custom usernames down below.
701        if allowed.contains(git2::CredentialType::SSH_KEY) && !tried_sshkey {
702            // If ssh-agent authentication fails, libgit2 will keep
703            // calling this callback asking for other authentication
704            // methods to try. Make sure we only try ssh-agent once,
705            // to avoid looping forever.
706            tried_sshkey = true;
707            let username = username.unwrap();
708            debug_assert!(!ssh_username_requested);
709            ssh_agent_attempts.push(username.to_string());
710            return git2::Cred::ssh_key_from_agent(username);
711        }
712
713        // Sometimes libgit2 will ask for a username/password in plaintext. This
714        // is where Cargo would have an interactive prompt if we supported it,
715        // but we currently don't! Right now the only way we support fetching a
716        // plaintext password is through the `credential.helper` support, so
717        // fetch that here.
718        //
719        // If ssh-agent authentication fails, libgit2 will keep calling this
720        // callback asking for other authentication methods to try. Check
721        // cred_helper_bad to make sure we only try the git credential helper
722        // once, to avoid looping forever.
723        if allowed.contains(git2::CredentialType::USER_PASS_PLAINTEXT) && cred_helper_bad.is_none()
724        {
725            let r = git2::Cred::credential_helper(cfg, url, username);
726            cred_helper_bad = Some(r.is_err());
727            return r;
728        }
729
730        // I'm... not sure what the DEFAULT kind of authentication is, but seems
731        // easy to support?
732        if allowed.contains(git2::CredentialType::DEFAULT) {
733            return git2::Cred::default();
734        }
735
736        // Whelp, we tried our best
737        Err(git2::Error::from_str("no authentication methods succeeded"))
738    });
739
740    // Ok, so if it looks like we're going to be doing ssh authentication, we
741    // want to try a few different usernames as one wasn't specified in the URL
742    // for us to use. In order, we'll try:
743    //
744    // * A credential helper's username for this URL, if available.
745    // * This account's username.
746    // * "git"
747    //
748    // We have to restart the authentication session each time (due to
749    // constraints in libssh2 I guess? maybe this is inherent to ssh?), so we
750    // call our callback, `f`, in a loop here.
751    if ssh_username_requested {
752        debug_assert!(res.is_err());
753        let mut attempts = vec![String::from("git")];
754        if let Ok(s) = gctx.get_env("USER").or_else(|_| gctx.get_env("USERNAME")) {
755            attempts.push(s.to_string());
756        }
757        if let Some(ref s) = cred_helper.username {
758            attempts.push(s.clone());
759        }
760
761        while let Some(s) = attempts.pop() {
762            // We should get `USERNAME` first, where we just return our attempt,
763            // and then after that we should get `SSH_KEY`. If the first attempt
764            // fails we'll get called again, but we don't have another option so
765            // we bail out.
766            let mut attempts = 0;
767            res = f(&mut |_url, username, allowed| {
768                if allowed.contains(git2::CredentialType::USERNAME) {
769                    return git2::Cred::username(&s);
770                }
771                if allowed.contains(git2::CredentialType::SSH_KEY) {
772                    debug_assert_eq!(Some(&s[..]), username);
773                    attempts += 1;
774                    if attempts == 1 {
775                        ssh_agent_attempts.push(s.to_string());
776                        return git2::Cred::ssh_key_from_agent(&s);
777                    }
778                }
779                Err(git2::Error::from_str("no authentication methods succeeded"))
780            });
781
782            // If we made two attempts then that means:
783            //
784            // 1. A username was requested, we returned `s`.
785            // 2. An ssh key was requested, we returned to look up `s` in the
786            //    ssh agent.
787            // 3. For whatever reason that lookup failed, so we were asked again
788            //    for another mode of authentication.
789            //
790            // Essentially, if `attempts == 2` then in theory the only error was
791            // that this username failed to authenticate (e.g., no other network
792            // errors happened). Otherwise something else is funny so we bail
793            // out.
794            if attempts != 2 {
795                break;
796            }
797        }
798    }
799    let mut err = match res {
800        Ok(e) => return Ok(e),
801        Err(e) => e,
802    };
803
804    // In the case of an authentication failure (where we tried something) then
805    // we try to give a more helpful error message about precisely what we
806    // tried.
807    if any_attempts {
808        let mut msg = "failed to authenticate when downloading \
809                       repository"
810            .to_string();
811
812        if let Some(attempt) = &url_attempt {
813            if url != attempt {
814                msg.push_str(": ");
815                msg.push_str(attempt);
816            }
817        }
818        msg.push('\n');
819        if !ssh_agent_attempts.is_empty() {
820            let names = ssh_agent_attempts
821                .iter()
822                .map(|s| format!("`{}`", s))
823                .collect::<Vec<_>>()
824                .join(", ");
825            msg.push_str(&format!(
826                "\n* attempted ssh-agent authentication, but \
827                 no usernames succeeded: {}",
828                names
829            ));
830        }
831        if let Some(failed_cred_helper) = cred_helper_bad {
832            if failed_cred_helper {
833                msg.push_str(
834                    "\n* attempted to find username/password via \
835                     git's `credential.helper` support, but failed",
836                );
837            } else {
838                msg.push_str(
839                    "\n* attempted to find username/password via \
840                     `credential.helper`, but maybe the found \
841                     credentials were incorrect",
842                );
843            }
844        }
845        msg.push_str("\n\n");
846        msg.push_str("if the git CLI succeeds then `net.git-fetch-with-cli` may help here\n");
847        msg.push_str("https://doc.rust-lang.org/cargo/reference/config.html#netgit-fetch-with-cli");
848        err = err.context(msg);
849
850        // Otherwise if we didn't even get to the authentication phase them we may
851        // have failed to set up a connection, in these cases hint on the
852        // `net.git-fetch-with-cli` configuration option.
853    } else if let Some(e) = err.downcast_ref::<git2::Error>() {
854        match e.class() {
855            ErrorClass::Net
856            | ErrorClass::Ssl
857            | ErrorClass::Submodule
858            | ErrorClass::FetchHead
859            | ErrorClass::Ssh
860            | ErrorClass::Http => {
861                let msg = format!(
862                    concat!(
863                        "network failure seems to have happened\n",
864                        "if a proxy or similar is necessary `net.git-fetch-with-cli` may help here\n",
865                        "https://doc.rust-lang.org/cargo/reference/config.html#netgit-fetch-with-cli",
866                        "{}"
867                    ),
868                    note_github_pull_request(url).unwrap_or_default()
869                );
870                err = err.context(msg);
871            }
872            ErrorClass::Callback => {
873                // This unwraps the git2 error. We're using the callback error
874                // specifically to convey errors from Rust land through the C
875                // callback interface. We don't need the `; class=Callback
876                // (26)` that gets tacked on to the git2 error message.
877                err = anyhow::format_err!("{}", e.message());
878            }
879            _ => {}
880        }
881    }
882
883    Err(err)
884}
885
886/// `git reset --hard` to the given `obj` for the `repo`.
887///
888/// The `obj` is a commit-ish to which the head should be moved.
889#[tracing::instrument(skip_all)]
890fn reset(repo: &git2::Repository, obj: &git2::Object<'_>, gctx: &GlobalContext) -> CargoResult<()> {
891    let mut pb = Progress::new("Checkout", gctx);
892    let mut opts = git2::build::CheckoutBuilder::new();
893    opts.progress(|_, cur, max| {
894        drop(pb.tick(cur, max, ""));
895    });
896    debug!("doing reset");
897    repo.reset(obj, git2::ResetType::Hard, Some(&mut opts))?;
898    debug!("reset done");
899    Ok(())
900}
901
902/// Prepares the callbacks for fetching a git repository.
903///
904/// The main purpose of this function is to construct everything before a fetch.
905/// This will attempt to setup a progress bar, the authentication for git,
906/// ssh known hosts check, and the network retry mechanism.
907///
908/// The callback is provided a fetch options, which can be used by the actual
909/// git fetch.
910pub fn with_fetch_options(
911    git_config: &git2::Config,
912    url: &str,
913    gctx: &GlobalContext,
914    cb: &mut dyn FnMut(git2::FetchOptions<'_>) -> CargoResult<()>,
915) -> CargoResult<()> {
916    let mut progress = Progress::new("Fetch", gctx);
917    let ssh_config = gctx.net_config()?.ssh.as_ref();
918    let config_known_hosts = ssh_config.and_then(|ssh| ssh.known_hosts.as_ref());
919    let diagnostic_home_config = gctx.diagnostic_home_config();
920    network::retry::with_retry(gctx, || {
921        // Hack: libgit2 disallows overriding the error from check_cb since v1.8.0,
922        // so we store the error additionally and unwrap it later
923        let mut check_cb_result = Ok(());
924        let auth_result = with_authentication(gctx, url, git_config, |f| {
925            let port = Url::parse(url).ok().and_then(|url| url.port());
926            let mut last_update = Instant::now();
927            let mut rcb = git2::RemoteCallbacks::new();
928            // We choose `N=10` here to make a `300ms * 10slots ~= 3000ms`
929            // sliding window for tracking the data transfer rate (in bytes/s).
930            let mut counter = MetricsCounter::<10>::new(0, last_update);
931            rcb.credentials(f);
932            rcb.certificate_check(|cert, host| {
933                match super::known_hosts::certificate_check(
934                    gctx,
935                    cert,
936                    host,
937                    port,
938                    config_known_hosts,
939                    &diagnostic_home_config,
940                ) {
941                    Ok(status) => Ok(status),
942                    Err(e) => {
943                        check_cb_result = Err(e);
944                        // This is not really used because it'll be overridden by libgit2
945                        // See https://github.com/libgit2/libgit2/commit/9a9f220119d9647a352867b24b0556195cb26548
946                        Err(git2::Error::from_str(
947                            "invalid or unknown remote ssh hostkey",
948                        ))
949                    }
950                }
951            });
952            rcb.transfer_progress(|stats| {
953                let indexed_deltas = stats.indexed_deltas();
954                let msg = if indexed_deltas > 0 {
955                    // Resolving deltas.
956                    format!(
957                        ", ({}/{}) resolving deltas",
958                        indexed_deltas,
959                        stats.total_deltas()
960                    )
961                } else {
962                    // Receiving objects.
963                    //
964                    // # Caveat
965                    //
966                    // Progress bar relies on git2 calling `transfer_progress`
967                    // to update its transfer rate, but we cannot guarantee a
968                    // periodic call of that callback. Thus if we don't receive
969                    // any data for, say, 10 seconds, the rate will get stuck
970                    // and never go down to 0B/s.
971                    // In the future, we need to find away to update the rate
972                    // even when the callback is not called.
973                    let now = Instant::now();
974                    // Scrape a `received_bytes` to the counter every 300ms.
975                    if now - last_update > Duration::from_millis(300) {
976                        counter.add(stats.received_bytes(), now);
977                        last_update = now;
978                    }
979                    let rate = HumanBytes(counter.rate() as u64);
980                    format!(", {rate:.2}/s")
981                };
982                progress
983                    .tick(stats.indexed_objects(), stats.total_objects(), &msg)
984                    .is_ok()
985            });
986
987            // Create a local anonymous remote in the repository to fetch the
988            // url
989            let mut opts = git2::FetchOptions::new();
990            opts.remote_callbacks(rcb);
991            cb(opts)
992        });
993        if auth_result.is_err() {
994            check_cb_result?;
995        }
996        auth_result?;
997        Ok(())
998    })
999}
1000
1001/// Attempts to fetch the given git `reference` for a Git repository.
1002///
1003/// This is the main entry for git clone/fetch. It does the followings:
1004///
1005/// * Turns [`GitReference`] into refspecs accordingly.
1006/// * Dispatches `git fetch` using libgit2, gitoxide, or git CLI.
1007///
1008/// The `remote_url` argument is the git remote URL where we want to fetch from.
1009///
1010/// The `remote_kind` argument is a thing for [`-Zgitoxide`] shallow clones
1011/// at this time. It could be extended when libgit2 supports shallow clones.
1012///
1013/// [`-Zgitoxide`]: https://doc.rust-lang.org/nightly/cargo/reference/unstable.html#gitoxide
1014#[tracing::instrument(skip_all)]
1015pub fn fetch(
1016    repo: &mut git2::Repository,
1017    remote_url: &str,
1018    manifest_reference: &GitReference,
1019    locked_reference: &GitReference,
1020    gctx: &GlobalContext,
1021    remote_kind: RemoteKind,
1022) -> CargoResult<()> {
1023    if let Some(offline_flag) = gctx.offline_flag() {
1024        anyhow::bail!(
1025            "attempting to update a git repository, but {offline_flag} \
1026             was specified"
1027        )
1028    }
1029
1030    let shallow = remote_kind.to_shallow_setting(repo.is_shallow(), gctx);
1031
1032    // Flag to keep track if the rev is a full commit hash
1033    let mut fast_path_rev: bool = false;
1034
1035    let oid_to_fetch = match github_fast_path(repo, remote_url, locked_reference, gctx) {
1036        Ok(FastPathRev::UpToDate) => return Ok(()),
1037        Ok(FastPathRev::NeedsFetch(rev)) => Some(rev),
1038        Ok(FastPathRev::Indeterminate) => None,
1039        Err(e) => {
1040            debug!("failed to check github {:?}", e);
1041            None
1042        }
1043    };
1044
1045    maybe_gc_repo(repo, gctx)?;
1046
1047    clean_repo_temp_files(repo);
1048
1049    // Translate the reference desired here into an actual list of refspecs
1050    // which need to get fetched. Additionally record if we're fetching tags.
1051    let mut refspecs = Vec::new();
1052    let mut tags = false;
1053    // The `+` symbol on the refspec means to allow a forced (fast-forward)
1054    // update which is needed if there is ever a force push that requires a
1055    // fast-forward.
1056    match locked_reference {
1057        // For branches and tags we can fetch simply one reference and copy it
1058        // locally, no need to fetch other branches/tags.
1059        GitReference::Branch(b) => {
1060            refspecs.push(format!("+refs/heads/{0}:refs/remotes/origin/{0}", b));
1061        }
1062
1063        GitReference::Tag(t) => {
1064            refspecs.push(format!("+refs/tags/{0}:refs/remotes/origin/tags/{0}", t));
1065        }
1066
1067        GitReference::DefaultBranch => {
1068            refspecs.push(String::from("+HEAD:refs/remotes/origin/HEAD"));
1069        }
1070
1071        GitReference::Rev(rev) => {
1072            if rev.starts_with("refs/") {
1073                refspecs.push(format!("+{0}:{0}", rev));
1074            } else if let Some(oid_to_fetch) = oid_to_fetch {
1075                fast_path_rev = true;
1076                refspecs.push(format!("+{0}:refs/commit/{0}", oid_to_fetch));
1077            } else if !matches!(shallow, gix::remote::fetch::Shallow::NoChange)
1078                && rev_to_oid(rev).is_some()
1079            {
1080                // There is a specific commit to fetch and we will do so in shallow-mode only
1081                // to not disturb the previous logic.
1082                // Note that with typical settings for shallowing, we will just fetch a single `rev`
1083                // as single commit.
1084                // The reason we write to `refs/remotes/origin/HEAD` is that it's of special significance
1085                // when during `GitReference::resolve()`, but otherwise it shouldn't matter.
1086                refspecs.push(format!("+{0}:refs/remotes/origin/HEAD", rev));
1087            } else if let GitReference::Rev(rev) = manifest_reference
1088                && rev.starts_with("refs/")
1089            {
1090                // If the lockfile has a commit. we can't directly fetch it (unless we're talking
1091                // to GitHub), so we fetch the ref associated with it from the manifest.
1092                refspecs.push(format!("+{0}:{0}", rev));
1093            } else {
1094                // We don't know what the rev will point to. To handle this
1095                // situation we fetch all branches and tags, and then we pray
1096                // it's somewhere in there.
1097                refspecs.push(String::from("+refs/heads/*:refs/remotes/origin/*"));
1098                refspecs.push(String::from("+HEAD:refs/remotes/origin/HEAD"));
1099                tags = true;
1100            }
1101        }
1102    }
1103
1104    debug!("doing a fetch for {remote_url}");
1105    let result = if let Some(true) = gctx.net_config()?.git_fetch_with_cli {
1106        fetch_with_cli(repo, remote_url, &refspecs, tags, shallow, gctx)
1107    } else if gctx.cli_unstable().gitoxide.map_or(false, |git| git.fetch) {
1108        fetch_with_gitoxide(repo, remote_url, refspecs, tags, shallow, gctx)
1109    } else {
1110        fetch_with_libgit2(repo, remote_url, refspecs, tags, shallow, gctx)
1111    };
1112
1113    if fast_path_rev {
1114        if let Some(oid) = oid_to_fetch {
1115            return result.with_context(|| format!("revision {} not found", oid));
1116        }
1117    }
1118    result
1119}
1120
1121/// `gitoxide` uses shallow locks to assure consistency when fetching to and to avoid races, and to write
1122/// files atomically.
1123/// Cargo has its own lock files and doesn't need that mechanism for race protection, so a stray lock means
1124/// a signal interrupted a previous shallow fetch and doesn't mean a race is happening.
1125fn has_shallow_lock_file(err: &crate::sources::git::fetch::Error) -> bool {
1126    matches!(
1127        err,
1128        gix::env::collate::fetch::Error::Fetch(gix::remote::fetch::Error::Fetch(
1129            gix::protocol::fetch::Error::LockShallowFile(_)
1130        ))
1131    )
1132}
1133
1134fn git_version() -> Option<GitVersion> {
1135    #[tracing::instrument(skip_all)]
1136    fn git_version() -> Option<GitVersion> {
1137        use std::process::Stdio;
1138
1139        let output = std::process::Command::new("git")
1140            .arg("--version")
1141            .stdin(Stdio::null())
1142            .output()
1143            .ok()?;
1144        if !output.status.success() {
1145            return None;
1146        }
1147        let Ok(stdout) = String::from_utf8(output.stdout) else {
1148            return None;
1149        };
1150        // All else fails, we can at least report the lowest common denominator of what we support
1151        let default_version = GitVersion {
1152            major: 2,
1153            minor: 0,
1154            patch: 0,
1155        };
1156        Some(GitVersion::from_version_stdout(&stdout).unwrap_or(default_version))
1157    }
1158
1159    static CACHE: std::sync::OnceLock<Option<GitVersion>> = std::sync::OnceLock::new();
1160    *CACHE.get_or_init(git_version)
1161}
1162
1163#[derive(Copy, Clone, Debug, PartialEq, Eq, PartialOrd, Ord)]
1164#[allow(unused)]
1165struct GitVersion {
1166    major: usize,
1167    minor: usize,
1168    patch: usize,
1169}
1170
1171impl GitVersion {
1172    fn from_version_stdout(stdout: &str) -> Result<Self, anyhow::Error> {
1173        // See https://github.com/git/git/blob/f78ce2f7b6df702f93d40b85d6bda92a3f65da79/help.c#L779-L784
1174        let Some(version) = stdout.strip_prefix("git version ") else {
1175            anyhow::bail!("unrecognized `git --version` output: {stdout}")
1176        };
1177        let (version, _) = version.split_once(" ").unwrap_or((version, ""));
1178        let (version, _) = version.split_once("\n").unwrap_or((version, ""));
1179        version.parse()
1180    }
1181}
1182
1183impl std::str::FromStr for GitVersion {
1184    type Err = anyhow::Error;
1185
1186    fn from_str(version: &str) -> Result<Self, Self::Err> {
1187        let unreleased = "GIT";
1188
1189        let s = version;
1190        let (major, s) = s.split_once(".").unwrap_or((s, ""));
1191        let mut major: usize = major.parse().map_err(|_err| {
1192            anyhow::format_err!("unrecognized major version `{major}` in `{version}`")
1193        })?;
1194        let (minor, s) = s.split_once(".").unwrap_or((s, ""));
1195        let mut minor: usize = if minor == unreleased {
1196            0
1197        } else {
1198            minor.parse().map_err(|_err| {
1199                anyhow::format_err!("unrecognized minor version `{minor}` in `{version}`")
1200            })?
1201        };
1202        let (patch, s) = s.split_once(".").unwrap_or((s, ""));
1203        let mut patch: usize = if patch == unreleased {
1204            0
1205        } else {
1206            patch.parse().map_err(|_err| {
1207                anyhow::format_err!("unrecognized patch version `{patch}` in `{version}`")
1208            })?
1209        };
1210
1211        let more = s;
1212        if !more.is_empty() {
1213            // `more` can be either pre-release or post-release.
1214            // The difference should be minimal, so be conservative and assume they are pre-release
1215            if let Some(sub) = patch.checked_sub(1) {
1216                patch = sub;
1217            } else if let Some(sub) = minor.checked_sub(1) {
1218                minor = sub;
1219            } else if let Some(sub) = major.checked_sub(1) {
1220                major = sub;
1221            }
1222        }
1223
1224        Ok(Self {
1225            major,
1226            minor,
1227            patch,
1228        })
1229    }
1230}
1231
1232/// Attempts to use `git` CLI installed on the system to fetch a repository,
1233/// when the config value [`net.git-fetch-with-cli`][1] is set.
1234///
1235/// Unfortunately `libgit2` is notably lacking in the realm of authentication
1236/// when compared to the `git` command line. As a result, allow an escape
1237/// hatch for users that would prefer to use `git`-the-CLI for fetching
1238/// repositories instead of `libgit2`-the-library. This should make more
1239/// flavors of authentication possible while also still giving us all the
1240/// speed and portability of using `libgit2`.
1241///
1242/// [1]: https://doc.rust-lang.org/nightly/cargo/reference/config.html#netgit-fetch-with-cli
1243#[tracing::instrument(skip(repo, gctx))]
1244fn fetch_with_cli(
1245    repo: &git2::Repository,
1246    url: &str,
1247    refspecs: &[String],
1248    tags: bool,
1249    shallow: gix::remote::fetch::Shallow,
1250    gctx: &GlobalContext,
1251) -> CargoResult<()> {
1252    debug!(target: "git-fetch", backend = "git-cli");
1253
1254    let Some(git_version) = git_version() else {
1255        anyhow::bail!(
1256            "`git` is not available
1257help: to still use `git` for fetching, please install it
1258help: to use Cargo's native git support, re-try with `net.git-fetch-with-cli = false`
1259https://doc.rust-lang.org/cargo/reference/config.html#netgit-fetch-with-cli"
1260        );
1261    };
1262
1263    let mut cmd = ProcessBuilder::new("git");
1264    // Avoid potential for unused work that may also hang (#15775)
1265    cmd.arg("-c").arg("core.fsmonitor=false");
1266    // Avoid warnings with `--no-show-forced-updates`
1267    cmd.arg("-c").arg("advice.fetchShowForcedUpdates=false");
1268    // Have 429's retried in git, where supported
1269    let max_retries = gctx
1270        .net_config()?
1271        .retry
1272        .unwrap_or(network::retry::MAX_RETRY_DEFAULT);
1273    cmd.arg("-c").arg(format!("http.maxRetries={max_retries}"));
1274    cmd.arg("-c").arg("http.retryAfter=1"); // `network::retry` has a growing value starting with 500ms, so choosing 1s
1275    cmd.arg("-c").arg(format!(
1276        "http.maxRetryTime={}",
1277        network::retry::MAX_RETRY_SLEEP_S
1278    ));
1279
1280    cmd.arg("fetch");
1281    if tags {
1282        cmd.arg("--tags");
1283    } else {
1284        cmd.arg("--no-tags");
1285    }
1286    if let gix::remote::fetch::Shallow::DepthAtRemote(depth) = shallow {
1287        let depth = 0i32.saturating_add_unsigned(depth.get());
1288        cmd.arg(format!("--depth={depth}"));
1289    }
1290
1291    let progress_config = gctx.progress_config();
1292    let progress = match progress_config.when {
1293        ProgressWhen::Always => true,
1294        ProgressWhen::Never => false,
1295        ProgressWhen::Auto => {
1296            // Recreate the same conditions used with `Progress`
1297            let width = progress_config
1298                .width
1299                .or_else(|| gctx.shell().err_width().progress_max_width());
1300            gctx.shell().progress_supported() && width.is_some()
1301        }
1302    };
1303    if gctx.shell().verbosity() == Verbosity::Verbose {
1304        cmd.arg("--verbose");
1305    } else if progress {
1306        cmd.arg("--progress");
1307        // https://github.com/git/git/blob/1630431f326e15fcde608827b5ff38422528eb59/Documentation/RelNotes/2.41.0.adoc?plain=1#L113-L114
1308        let min_version_porcelain = GitVersion {
1309            major: 2,
1310            minor: 41,
1311            patch: 0,
1312        };
1313        if min_version_porcelain <= git_version {
1314            // Move ref update status to `stdout` and silence it
1315            cmd.arg("--porcelain").stdout(Stdio::Null);
1316        }
1317    } else {
1318        cmd.arg("--quiet");
1319    }
1320
1321    // https://github.com/git/git/blob/1630431f326e15fcde608827b5ff38422528eb59/Documentation/RelNotes/2.23.0.adoc?plain=1#L56-L59
1322    let min_version_no_show_forced_update = GitVersion {
1323        major: 2,
1324        minor: 23,
1325        patch: 0,
1326    };
1327    if min_version_no_show_forced_update <= git_version {
1328        // skip unneeded expensive calculations
1329        cmd.arg("--no-show-forced-updates");
1330    }
1331
1332    cmd.arg("--force") // handle force pushes
1333        .arg("--update-head-ok") // see discussion in #2078
1334        .arg("--recurse-submodules=no") // we handle this, incompatible with `--porcelain`
1335        .arg(url)
1336        .args(refspecs)
1337        // If cargo is run by git (for example, the `exec` command in `git
1338        // rebase`), the GIT_DIR is set by git and will point to the wrong
1339        // location. This makes sure GIT_DIR is always the repository path.
1340        .env("GIT_DIR", repo.path())
1341        // The reset of these may not be necessary, but I'm including them
1342        // just to be extra paranoid and avoid any issues.
1343        .env_remove("GIT_WORK_TREE")
1344        .env_remove("GIT_INDEX_FILE")
1345        .env_remove("GIT_OBJECT_DIRECTORY")
1346        .env_remove("GIT_ALTERNATE_OBJECT_DIRECTORIES")
1347        .cwd(repo.path());
1348    gctx.shell()
1349        .verbose(|s| s.status("Running", &cmd.to_string()))?;
1350    // https://github.com/git/git/blob/1630431f326e15fcde608827b5ff38422528eb59/Documentation/RelNotes/2.54.0.adoc?plain=1#L107
1351    let min_version_max_retries = GitVersion {
1352        major: 2,
1353        minor: 54,
1354        patch: 0,
1355    };
1356    network::retry::with_retry(gctx, || {
1357        cmd.exec().map_err(|_error| {
1358            let spurious = git_version < min_version_max_retries;
1359            let pr_hint = note_github_pull_request(url).unwrap_or_default();
1360            let ssh_config_hint = note_ssh_config(gctx).unwrap_or_default().unwrap_or_default();
1361            let git_config_hint = note_git_config(gctx).unwrap_or_default().unwrap_or_default();
1362            let with_depth = if let gix::remote::fetch::Shallow::DepthAtRemote(depth) = shallow {
1363                format!(" with depth={depth}")
1364            } else {
1365                "".to_owned()
1366            };
1367            GitCliError::new(anyhow::format_err!("`git fetch` failed for {url}{with_depth}{pr_hint}{ssh_config_hint}{git_config_hint}"))
1368                .spurious(spurious)
1369                .workaround(
1370                    "help: re-try with `net.git-fetch-with-cli = false` to see if it resolves the problem
1371https://doc.rust-lang.org/cargo/reference/config.html#netgit-fetch-with-cli",
1372                )
1373                .into()
1374        })
1375    })?;
1376
1377    Ok(())
1378}
1379
1380#[tracing::instrument(skip(repo, gctx))]
1381fn fetch_with_gitoxide(
1382    repo: &mut git2::Repository,
1383    remote_url: &str,
1384    refspecs: Vec<String>,
1385    tags: bool,
1386    shallow: gix::remote::fetch::Shallow,
1387    gctx: &GlobalContext,
1388) -> CargoResult<()> {
1389    debug!(target: "git-fetch", backend = "gitoxide");
1390
1391    let git2_repo = repo;
1392    let config_overrides = cargo_config_to_gitoxide_overrides(gctx)?;
1393    let repo_reinitialized = AtomicBool::default();
1394    let res = oxide::with_retry_and_progress(
1395        git2_repo.path(),
1396        gctx,
1397        remote_url,
1398        &|repo_path,
1399          should_interrupt,
1400          mut progress,
1401          url_for_authentication: &mut dyn FnMut(&gix::bstr::BStr)| {
1402            // The `fetch` operation here may fail spuriously due to a corrupt
1403            // repository. It could also fail, however, for a whole slew of other
1404            // reasons (aka network related reasons). We want Cargo to automatically
1405            // recover from corrupt repositories, but we don't want Cargo to stomp
1406            // over other legitimate errors.
1407            //
1408            // Consequently we save off the error of the `fetch` operation and if it
1409            // looks like a "corrupt repo" error then we blow away the repo and try
1410            // again. If it looks like any other kind of error, or if we've already
1411            // blown away the repository, then we want to return the error as-is.
1412            loop {
1413                let res = oxide::open_repo(
1414                    repo_path,
1415                    config_overrides.clone(),
1416                    oxide::OpenMode::ForFetch,
1417                )
1418                .map_err(crate::sources::git::fetch::Error::from)
1419                .and_then(|repo| {
1420                    debug!("initiating fetch of {refspecs:?} from {remote_url}");
1421                    let url_for_authentication = &mut *url_for_authentication;
1422                    let remote = repo
1423                        .remote_at(remote_url)?
1424                        .with_fetch_tags(if tags {
1425                            gix::remote::fetch::Tags::All
1426                        } else {
1427                            gix::remote::fetch::Tags::Included
1428                        })
1429                        .with_refspecs(
1430                            refspecs.iter().map(|s| s.as_str()),
1431                            gix::remote::Direction::Fetch,
1432                        )
1433                        .map_err(crate::sources::git::fetch::Error::Other)?;
1434                    let url = remote
1435                        .url(gix::remote::Direction::Fetch)
1436                        .expect("set at init")
1437                        .to_owned();
1438                    let connection = remote.connect(gix::remote::Direction::Fetch)?;
1439                    let mut authenticate = connection.configured_credentials(url)?;
1440                    let connection = connection.with_credentials(
1441                        move |action: gix::protocol::credentials::helper::Action| {
1442                            if let Some(url) = action
1443                                .context()
1444                                .and_then(|gctx| gctx.url.as_ref().filter(|url| *url != remote_url))
1445                            {
1446                                url_for_authentication(url.as_ref());
1447                            }
1448                            authenticate(action)
1449                        },
1450                    );
1451                    let outcome = connection
1452                        .prepare_fetch(&mut progress, gix::remote::ref_map::Options::default())?
1453                        .with_shallow(shallow.clone())
1454                        .receive(&mut progress, should_interrupt)?;
1455                    Ok(outcome)
1456                });
1457                let err = match res {
1458                    Ok(_) => break,
1459                    Err(e) => e,
1460                };
1461                debug!("fetch failed: {}", err);
1462
1463                if !repo_reinitialized.load(Ordering::Relaxed)
1464                        // We check for errors that could occur if the configuration, refs or odb files are corrupted.
1465                        // We don't check for errors related to writing as `gitoxide` is expected to create missing leading
1466                        // folder before writing files into it, or else not even open a directory as git repository (which is
1467                        // also handled here).
1468                        && err.is_corrupted()
1469                    || has_shallow_lock_file(&err)
1470                {
1471                    repo_reinitialized.store(true, Ordering::Relaxed);
1472                    debug!(
1473                        "looks like this is a corrupt repository, reinitializing \
1474                     and trying again"
1475                    );
1476                    if oxide::reinitialize(repo_path).is_ok() {
1477                        continue;
1478                    }
1479                }
1480
1481                return Err(err.into());
1482            }
1483            Ok(())
1484        },
1485    );
1486    if repo_reinitialized.load(Ordering::Relaxed) {
1487        *git2_repo = git2::Repository::open(git2_repo.path())?;
1488    }
1489    res
1490}
1491
1492#[tracing::instrument(skip(repo, gctx))]
1493fn fetch_with_libgit2(
1494    repo: &mut git2::Repository,
1495    remote_url: &str,
1496    refspecs: Vec<String>,
1497    tags: bool,
1498    shallow: gix::remote::fetch::Shallow,
1499    gctx: &GlobalContext,
1500) -> CargoResult<()> {
1501    debug!(target: "git-fetch", backend = "libgit2");
1502
1503    static INIT: Once = Once::new();
1504    INIT.call_once(|| {
1505        // SAFETY:
1506        //
1507        // The unsafety of the registration function derives from two aspects:
1508        //
1509        // 1. This call must be synchronized with all other registration calls as
1510        //    well as construction of new transports.
1511        // 2. The argument is leaked.
1512        //
1513        // We're clear on point (1) because this is the only `register` call we make in the
1514        // cargo-binary and cargo-library is not officially supported and it is under a lock.
1515        // Technically, `git2_curl` also has a lock but that isn't part of their API guarantees.
1516        //
1517        // We're mostly clear on point (2) because we'd only free it after everything is done anyway
1518        unsafe { init_git_transports(gctx) };
1519    });
1520
1521    let git_config = git2::Config::open_default()?;
1522    with_fetch_options(&git_config, remote_url, gctx, &mut |mut opts| {
1523        if tags {
1524            opts.download_tags(git2::AutotagOption::All);
1525        }
1526        if let gix::remote::fetch::Shallow::DepthAtRemote(depth) = shallow {
1527            opts.depth(0i32.saturating_add_unsigned(depth.get()));
1528        }
1529        // The `fetch` operation here may fail spuriously due to a corrupt
1530        // repository. It could also fail, however, for a whole slew of other
1531        // reasons (aka network related reasons). We want Cargo to automatically
1532        // recover from corrupt repositories, but we don't want Cargo to stomp
1533        // over other legitimate errors.
1534        //
1535        // Consequently we save off the error of the `fetch` operation and if it
1536        // looks like a "corrupt repo" error then we blow away the repo and try
1537        // again. If it looks like any other kind of error, or if we've already
1538        // blown away the repository, then we want to return the error as-is.
1539        let mut repo_reinitialized = false;
1540        loop {
1541            debug!("initiating fetch of {refspecs:?} from {remote_url}");
1542            let res = repo
1543                .remote_anonymous(remote_url)?
1544                .fetch(&refspecs, Some(&mut opts), None);
1545            let err = match res {
1546                Ok(()) => break,
1547                Err(e) => e,
1548            };
1549            debug!("fetch failed: {}", err);
1550
1551            if !repo_reinitialized && matches!(err.class(), ErrorClass::Reference | ErrorClass::Odb)
1552            {
1553                repo_reinitialized = true;
1554                debug!(
1555                    "looks like this is a corrupt repository, reinitializing \
1556                     and trying again"
1557                );
1558                if reinitialize(repo).is_ok() {
1559                    continue;
1560                }
1561            }
1562
1563            return Err(err.into());
1564        }
1565        Ok(())
1566    })
1567}
1568
1569/// Configure libgit2 to use libcurl if necessary.
1570///
1571/// If the user has a non-default network configuration, then libgit2 will be
1572/// configured to use libcurl instead of the built-in networking support so
1573/// that those configuration settings can be used.
1574///
1575/// # Safety
1576///
1577/// See [git2_curl::register]
1578#[tracing::instrument(skip_all)]
1579unsafe fn init_git_transports(gctx: &GlobalContext) {
1580    match needs_custom_http_transport(gctx) {
1581        Ok(true) => {}
1582        _ => return,
1583    }
1584
1585    let handle = match http_handle(gctx) {
1586        Ok(handle) => handle,
1587        Err(..) => return,
1588    };
1589
1590    unsafe {
1591        git2_curl::register(handle);
1592    }
1593}
1594
1595/// Attempts to `git gc` a repository.
1596///
1597/// Cargo has a bunch of long-lived git repositories in its global cache and
1598/// some, like the index, are updated very frequently. Right now each update
1599/// creates a new "pack file" inside the git database, and over time this can
1600/// cause bad performance and bad current behavior in libgit2.
1601///
1602/// One pathological use case today is where libgit2 opens hundreds of file
1603/// descriptors, getting us dangerously close to blowing out the OS limits of
1604/// how many fds we can have open. This is detailed in [#4403].
1605///
1606/// Instead of trying to be clever about when gc is needed, we just run
1607/// `git gc --auto` and let git figure it out. It checks its own thresholds
1608/// (gc.auto, gc.autoPackLimit) and either does the work or exits quickly.
1609/// If git isn't installed, no worries - we skip it.
1610///
1611/// [#4403]: https://github.com/rust-lang/cargo/issues/4403
1612fn maybe_gc_repo(repo: &mut git2::Repository, gctx: &GlobalContext) -> CargoResult<()> {
1613    // Let git decide whether gc is actually needed based on its own thresholds
1614    // (gc.auto, gc.autoPackLimit). This avoids duplicating git's internal logic
1615    // for deciding when housekeeping is needed.
1616    //
1617    // For testing purposes, __CARGO_PACKFILE_LIMIT can be set to override
1618    // gc.autoPackLimit, which has the same meaning. This lets tests force gc
1619    // to run by setting a low threshold without depending on git's defaults.
1620    let mut cmd = Command::new("git");
1621    if let Ok(limit) = gctx.get_env("__CARGO_PACKFILE_LIMIT") {
1622        cmd.arg(format!("-c gc.autoPackLimit={}", limit));
1623    }
1624    cmd.arg("gc")
1625        .arg("--auto")
1626        // Explicitly set `GIT_DIR` so `safe.bareRepository=explicit` doesn't reject it.
1627        .env("GIT_DIR", repo.path())
1628        .current_dir(repo.path());
1629
1630    match cmd.output() {
1631        Ok(out) => {
1632            debug!(
1633                "git-gc --auto status: {}\n\nstdout ---\n{}\nstderr ---\n{}",
1634                out.status,
1635                String::from_utf8_lossy(&out.stdout),
1636                String::from_utf8_lossy(&out.stderr)
1637            );
1638            if out.status.success() {
1639                let new = git2::Repository::open(repo.path())?;
1640                *repo = new;
1641                return Ok(());
1642            }
1643        }
1644        Err(e) => debug!("git-gc --auto failed to spawn: {}", e),
1645    }
1646
1647    // Alright all else failed, let's start over.
1648    reinitialize(repo)
1649}
1650
1651/// Removes temporary files left from previous activity.
1652///
1653/// If libgit2 is interrupted while indexing pack files, it will leave behind
1654/// some temporary files that it doesn't clean up. These can be quite large in
1655/// size, so this tries to clean things up.
1656///
1657/// This intentionally ignores errors. This is only an opportunistic cleaning,
1658/// and we don't really care if there are issues (there's unlikely anything
1659/// that can be done).
1660///
1661/// The git CLI has similar behavior (its temp files look like
1662/// `objects/pack/tmp_pack_9kUSA8`). Those files are normally deleted via `git
1663/// prune` which is run by `git gc`. However, it doesn't know about libgit2's
1664/// filenames, so they never get cleaned up.
1665fn clean_repo_temp_files(repo: &git2::Repository) {
1666    let path = repo.path().join("objects/pack/pack_git2_*");
1667    let Some(pattern) = path.to_str() else {
1668        tracing::warn!("cannot convert {path:?} to a string");
1669        return;
1670    };
1671    let Ok(paths) = glob::glob(pattern) else {
1672        return;
1673    };
1674    for path in paths {
1675        if let Ok(path) = path {
1676            match paths::remove_file(&path) {
1677                Ok(_) => tracing::debug!("removed stale temp git file {path:?}"),
1678                Err(e) => {
1679                    tracing::warn!("failed to remove {path:?} while cleaning temp files: {e}")
1680                }
1681            }
1682        }
1683    }
1684}
1685
1686/// Reinitializes a given Git repository. This is useful when a Git repository
1687/// seems corrupted and we want to start over.
1688fn reinitialize(repo: &mut git2::Repository) -> CargoResult<()> {
1689    // Here we want to drop the current repository object pointed to by `repo`,
1690    // so we initialize temporary repository in a sub-folder, blow away the
1691    // existing git folder, and then recreate the git repo. Finally we blow away
1692    // the `tmp` folder we allocated.
1693    let path = repo.path().to_path_buf();
1694    debug!("reinitializing git repo at {:?}", path);
1695    let tmp = path.join("tmp");
1696    let bare = !repo.path().ends_with(".git");
1697    *repo = init(&tmp, false)?;
1698    for entry in path.read_dir()? {
1699        let entry = entry?;
1700        if entry.file_name().to_str() == Some("tmp") {
1701            continue;
1702        }
1703        let path = entry.path();
1704        drop(paths::remove_file(&path).or_else(|_| paths::remove_dir_all(&path)));
1705    }
1706    *repo = init(&path, bare)?;
1707    paths::remove_dir_all(&tmp)?;
1708    Ok(())
1709}
1710
1711/// Initializes a Git repository at `path`.
1712fn init(path: &Path, bare: bool) -> CargoResult<git2::Repository> {
1713    let mut opts = git2::RepositoryInitOptions::new();
1714    // Skip anything related to templates, they just call all sorts of issues as
1715    // we really don't want to use them yet they insist on being used. See #6240
1716    // for an example issue that comes up.
1717    opts.external_template(false);
1718    opts.bare(bare);
1719    Ok(git2::Repository::init_opts(&path, &opts)?)
1720}
1721
1722/// The result of GitHub fast path check. See [`github_fast_path`] for more.
1723enum FastPathRev {
1724    /// The local rev (determined by `reference.resolve(repo)`) is already up to
1725    /// date with what this rev resolves to on GitHub's server.
1726    UpToDate,
1727    /// The following SHA must be fetched in order for the local rev to become
1728    /// up to date.
1729    NeedsFetch(Oid),
1730    /// Don't know whether local rev is up to date. We'll fetch _all_ branches
1731    /// and tags from the server and see what happens.
1732    Indeterminate,
1733}
1734
1735/// Attempts GitHub's special fast path for testing if we've already got an
1736/// up-to-date copy of the repository.
1737///
1738/// Updating the index is done pretty regularly so we want it to be as fast as
1739/// possible. For registries hosted on GitHub (like the crates.io index) there's
1740/// a fast path available to use[^1] to tell us that there's no updates to be
1741/// made.
1742///
1743/// Note that this function should never cause an actual failure because it's
1744/// just a fast path. As a result, a caller should ignore `Err` returned from
1745/// this function and move forward on the normal path.
1746///
1747/// [^1]: <https://developer.github.com/v3/repos/commits/#get-the-sha-1-of-a-commit-reference>
1748#[tracing::instrument(skip(repo, gctx))]
1749fn github_fast_path(
1750    repo: &git2::Repository,
1751    url: &str,
1752    reference: &GitReference,
1753    gctx: &GlobalContext,
1754) -> CargoResult<FastPathRev> {
1755    let url = Url::parse(url)?;
1756    if !is_github(&url) {
1757        return Ok(FastPathRev::Indeterminate);
1758    }
1759
1760    let local_object = resolve_ref(reference, repo).ok();
1761
1762    let github_branch_name = match reference {
1763        GitReference::Branch(branch) => branch,
1764        GitReference::Tag(tag) => tag,
1765        GitReference::DefaultBranch => "HEAD",
1766        GitReference::Rev(rev) => {
1767            if rev.starts_with("refs/") {
1768                rev
1769            } else if looks_like_commit_hash(rev) {
1770                // `revparse_single` (used by `resolve`) is the only way to turn
1771                // short hash -> long hash, but it also parses other things,
1772                // like branch and tag names, which might coincidentally be
1773                // valid hex.
1774                //
1775                // We only return early if `rev` is a prefix of the object found
1776                // by `revparse_single`. Don't bother talking to GitHub in that
1777                // case, since commit hashes are permanent. If a commit with the
1778                // requested hash is already present in the local clone, its
1779                // contents must be the same as what is on the server for that
1780                // hash.
1781                //
1782                // If `rev` is not found locally by `revparse_single`, we'll
1783                // need GitHub to resolve it and get a hash. If `rev` is found
1784                // but is not a short hash of the found object, it's probably a
1785                // branch and we also need to get a hash from GitHub, in case
1786                // the branch has moved.
1787                if let Some(local_object) = local_object {
1788                    if is_short_hash_of(rev, local_object) {
1789                        debug!("github fast path already has {local_object}");
1790                        return Ok(FastPathRev::UpToDate);
1791                    }
1792                }
1793                // If `rev` is a full commit hash, the only thing it can resolve
1794                // to is itself. Don't bother talking to GitHub in that case
1795                // either. (This ensures that we always attempt to fetch the
1796                // commit directly even if we can't reach the GitHub API.)
1797                if let Some(oid) = rev_to_oid(rev) {
1798                    debug!("github fast path is already a full commit hash {rev}");
1799                    return Ok(FastPathRev::NeedsFetch(oid));
1800                }
1801                rev
1802            } else {
1803                debug!("can't use github fast path with `rev = \"{}\"`", rev);
1804                return Ok(FastPathRev::Indeterminate);
1805            }
1806        }
1807    };
1808
1809    // This expects GitHub urls in the form `github.com/user/repo` and nothing
1810    // else
1811    let mut pieces = url
1812        .path_segments()
1813        .ok_or_else(|| anyhow!("no path segments on url"))?;
1814    let username = pieces
1815        .next()
1816        .ok_or_else(|| anyhow!("couldn't find username"))?;
1817    let repository = pieces
1818        .next()
1819        .ok_or_else(|| anyhow!("couldn't find repository name"))?;
1820    if pieces.next().is_some() {
1821        anyhow::bail!("too many segments on URL");
1822    }
1823
1824    // Trim off the `.git` from the repository, if present, since that's
1825    // optional for GitHub and won't work when we try to use the API as well.
1826    let repository = repository.strip_suffix(".git").unwrap_or(repository);
1827
1828    let url = format!(
1829        "https://api.github.com/repos/{}/{}/commits/{}",
1830        username, repository, github_branch_name,
1831    );
1832    debug!("attempting GitHub fast path for {}", url);
1833    let mut request =
1834        Request::get(url).header(http::header::ACCEPT, "application/vnd.github.3.sha");
1835    if let Some(local_object) = local_object {
1836        request = request.header(http::header::IF_NONE_MATCH, &format!("\"{local_object}\""));
1837    }
1838    let response = gctx
1839        .http_async()?
1840        .request_blocking(request.body(Vec::new())?)?;
1841    let response_code = response.status();
1842    if response_code == StatusCode::NOT_MODIFIED {
1843        debug!("github fast path up-to-date");
1844        Ok(FastPathRev::UpToDate)
1845    } else if response_code == StatusCode::OK
1846        && let Some(oid_to_fetch) = rev_to_oid(str::from_utf8(&response.body())?)
1847    {
1848        // response expected to be a full hash hexstring (40 or 64 chars)
1849        debug!("github fast path fetch {oid_to_fetch}");
1850        Ok(FastPathRev::NeedsFetch(oid_to_fetch))
1851    } else {
1852        // Usually response_code == 404 if the repository does not exist, and
1853        // response_code == 422 if exists but GitHub is unable to resolve the
1854        // requested rev.
1855        debug!("github fast path bad response code {response_code}");
1856        Ok(FastPathRev::Indeterminate)
1857    }
1858}
1859
1860/// Whether a `url` is one from GitHub.
1861fn is_github(url: &Url) -> bool {
1862    url.host_str() == Some("github.com")
1863}
1864
1865// Give some messages on GitHub PR URL given as is
1866pub(crate) fn note_github_pull_request(url: &str) -> Option<String> {
1867    if let Ok(url) = url.parse::<Url>()
1868        && is_github(&url)
1869    {
1870        let path_segments = url
1871            .path_segments()
1872            .map(|p| p.into_iter().collect::<Vec<_>>())
1873            .unwrap_or_default();
1874        if let [owner, repo, "pull", pr_number, ..] = path_segments[..] {
1875            let repo_url = format!("https://github.com/{owner}/{repo}.git");
1876            let rev = format!("refs/pull/{pr_number}/head");
1877            return Some(format!(
1878                concat!(
1879                    "\n\nnote: GitHub url {} is not a repository\n",
1880                    "help: to specify a pull request as a dependency, replace the dependency with:\n",
1881                    "       `git = \"{}\" rev = \"{}\"`",
1882                ),
1883                url, repo_url, rev
1884            ));
1885        }
1886    }
1887
1888    None
1889}
1890
1891pub(crate) fn note_ssh_config(gctx: &GlobalContext) -> CargoResult<Option<&'static str>> {
1892    let net_config = gctx.net_config()?;
1893    if let Some(ssh_config) = net_config.ssh.as_ref()
1894        && ssh_config.known_hosts.is_some()
1895    {
1896        Ok(Some(
1897            "
1898note: `cargo`s `net.ssh.known-hosts` is not applied to `git`, check your system SSH configuration to ensure it is set there",
1899        ))
1900    } else {
1901        Ok(None)
1902    }
1903}
1904
1905pub(crate) fn note_git_config(gctx: &GlobalContext) -> CargoResult<Option<String>> {
1906    let mut fields = Vec::new();
1907
1908    // As this is for errors, the focus is on things that impact functionality
1909    let http_config = gctx.http_config()?;
1910    if http_config.proxy.is_some() {
1911        fields.push("http.proxy (git: http.proxy)");
1912    }
1913    if http_config.low_speed_limit.is_some() {
1914        fields.push("http.low-speed-limit (git: http.lowSpeedLimit)");
1915    }
1916    if http_config.timeout.is_some() {
1917        fields.push("http.timeout (git: http.lowSpeedTime)");
1918    }
1919    if http_config.cainfo.is_some() {
1920        fields.push("http.cainfo (git: http.sslCAInfo)");
1921    }
1922    if http_config.proxy_cainfo.is_some() {
1923        fields.push("http.proxy-cainfo (git: http.proxySSLCAInfo)");
1924    }
1925    if http_config.check_revoke.is_some() {
1926        fields.push("http.check-revoke (git: http.schannelCheckRevoke)");
1927    }
1928    if http_config.multiplexing.is_some() {
1929        fields.push("http.multiplexing (git: http.version)");
1930    }
1931    if http_config.ssl_version.is_some() {
1932        fields.push("http.ssl-version (git: http.sslVersion)");
1933    }
1934
1935    if !fields.is_empty() {
1936        Ok(Some(format!("
1937note: `cargo` does not apply the following config to `git`, check your git configuration to ensure they are set there
1938    {}", fields.join("\n    "))))
1939    } else {
1940        Ok(None)
1941    }
1942}
1943
1944/// Whether a `rev` looks like a commit hash (ASCII hex digits).
1945fn looks_like_commit_hash(rev: &str) -> bool {
1946    rev.len() >= 7 && rev.chars().all(|ch| ch.is_ascii_hexdigit())
1947}
1948
1949/// Whether `rev` is a shorter hash of `oid`.
1950fn is_short_hash_of(rev: &str, oid: Oid) -> bool {
1951    let long_hash = oid.to_string();
1952    match long_hash.get(..rev.len()) {
1953        Some(truncated_long_hash) => truncated_long_hash.eq_ignore_ascii_case(rev),
1954        None => false,
1955    }
1956}
1957
1958#[cfg(test)]
1959mod tests {
1960    use super::*;
1961
1962    #[test]
1963    fn github_fast_path_full_hash_returns_needs_fetch() {
1964        let temp_dir = tempfile::TempDir::new().unwrap();
1965        let repo = git2::Repository::init_bare(temp_dir.path()).unwrap();
1966        let full_hash = "c9040898c9183ddbb9402dcbf749ed06d6ea90ad";
1967        let reference = GitReference::Rev(full_hash.to_string());
1968        let gctx = GlobalContext::default().unwrap();
1969        let expected_oid = rev_to_oid(full_hash).unwrap();
1970
1971        let result =
1972            github_fast_path(&repo, "https://github.com/user/repo", &reference, &gctx).unwrap();
1973
1974        assert!(matches!(result, FastPathRev::NeedsFetch(oid) if oid == expected_oid));
1975    }
1976
1977    #[test]
1978    fn test_absolute_submodule_url() {
1979        let cases = [
1980            (
1981                "ssh://git@gitub.com/rust-lang/cargo",
1982                "git@github.com:rust-lang/cargo.git",
1983                "git@github.com:rust-lang/cargo.git",
1984            ),
1985            (
1986                "ssh://git@gitub.com/rust-lang/cargo",
1987                "./",
1988                "ssh://git@gitub.com/rust-lang/cargo/",
1989            ),
1990            (
1991                "ssh://git@gitub.com/rust-lang/cargo",
1992                "../",
1993                "ssh://git@gitub.com/rust-lang/",
1994            ),
1995            (
1996                "ssh://git@gitub.com/rust-lang/cargo",
1997                "./foo",
1998                "ssh://git@gitub.com/rust-lang/cargo/foo",
1999            ),
2000            (
2001                "ssh://git@gitub.com/rust-lang/cargo/",
2002                "./foo",
2003                "ssh://git@gitub.com/rust-lang/cargo/foo",
2004            ),
2005            (
2006                "ssh://git@gitub.com/rust-lang/cargo/",
2007                "../foo",
2008                "ssh://git@gitub.com/rust-lang/foo",
2009            ),
2010            (
2011                "ssh://git@gitub.com/rust-lang/cargo",
2012                "../foo",
2013                "ssh://git@gitub.com/rust-lang/foo",
2014            ),
2015            (
2016                "ssh://git@gitub.com/rust-lang/cargo",
2017                "../foo/bar/../baz",
2018                "ssh://git@gitub.com/rust-lang/foo/baz",
2019            ),
2020            (
2021                "git@github.com:rust-lang/cargo.git",
2022                "ssh://git@gitub.com/rust-lang/cargo",
2023                "ssh://git@gitub.com/rust-lang/cargo",
2024            ),
2025            (
2026                "git@github.com:rust-lang/cargo.git",
2027                "./",
2028                "git@github.com:rust-lang/cargo.git/./",
2029            ),
2030            (
2031                "git@github.com:rust-lang/cargo.git",
2032                "../",
2033                "git@github.com:rust-lang/cargo.git/../",
2034            ),
2035            (
2036                "git@github.com:rust-lang/cargo.git",
2037                "./foo",
2038                "git@github.com:rust-lang/cargo.git/./foo",
2039            ),
2040            (
2041                "git@github.com:rust-lang/cargo.git/",
2042                "./foo",
2043                "git@github.com:rust-lang/cargo.git/./foo",
2044            ),
2045            (
2046                "git@github.com:rust-lang/cargo.git",
2047                "../foo",
2048                "git@github.com:rust-lang/cargo.git/../foo",
2049            ),
2050            (
2051                "git@github.com:rust-lang/cargo.git/",
2052                "../foo",
2053                "git@github.com:rust-lang/cargo.git/../foo",
2054            ),
2055            (
2056                "git@github.com:rust-lang/cargo.git",
2057                "../foo/bar/../baz",
2058                "git@github.com:rust-lang/cargo.git/../foo/bar/../baz",
2059            ),
2060        ];
2061
2062        for (base_url, submodule_url, expected) in cases {
2063            let url = absolute_submodule_url(base_url, submodule_url).unwrap();
2064            assert_eq!(
2065                expected, url,
2066                "base `{base_url}`; submodule `{submodule_url}`"
2067            );
2068        }
2069    }
2070}
2071
2072/// Turns a full commit hash revision into an oid.
2073///
2074/// Git object ID is supposed to be a hex string of 20 (SHA1) or 32 (SHA256) bytes.
2075/// Its length must be double to the underlying bytes (40 or 64),
2076/// otherwise libgit2 would happily zero-pad the returned oid.
2077///
2078/// See:
2079///
2080/// * <https://github.com/rust-lang/cargo/issues/13188>
2081/// * <https://github.com/rust-lang/cargo/issues/13968>
2082pub(super) fn rev_to_oid(rev: &str) -> Option<Oid> {
2083    Oid::from_str(rev)
2084        .ok()
2085        .filter(|oid| oid.as_bytes().len() * 2 == rev.len())
2086}
2087
2088#[cfg(test)]
2089mod test {
2090    use super::*;
2091
2092    #[test]
2093    fn git_version_empty() {
2094        let input = "";
2095        GitVersion::from_version_stdout(input).unwrap_err();
2096    }
2097
2098    #[test]
2099    fn git_version_version() {
2100        let input = "git version 2.19.0";
2101        let actual = GitVersion::from_version_stdout(input).unwrap();
2102        let expected = GitVersion {
2103            major: 2,
2104            minor: 19,
2105            patch: 0,
2106        };
2107        assert_eq!(actual, expected);
2108    }
2109
2110    #[test]
2111    fn git_version_version_nl() {
2112        let input = "git version 2.19.0\n";
2113        let actual = GitVersion::from_version_stdout(input).unwrap();
2114        let expected = GitVersion {
2115            major: 2,
2116            minor: 19,
2117            patch: 0,
2118        };
2119        assert_eq!(actual, expected);
2120    }
2121
2122    #[test]
2123    fn git_version_version_platform() {
2124        let input = "git version 2.19.0 (Apple Git-117)";
2125        let actual = GitVersion::from_version_stdout(input).unwrap();
2126        let expected = GitVersion {
2127            major: 2,
2128            minor: 19,
2129            patch: 0,
2130        };
2131        assert_eq!(actual, expected);
2132    }
2133
2134    #[test]
2135    fn git_version_version_vendor() {
2136        let input = "git version 2.19.0.windows.1";
2137        let actual = GitVersion::from_version_stdout(input).unwrap();
2138        let expected = GitVersion {
2139            major: 2,
2140            minor: 18,
2141            patch: 0,
2142        };
2143        assert_eq!(actual, expected);
2144    }
2145
2146    #[test]
2147    fn git_version_version_rc() {
2148        let input = "git version 2.19.0.rc0";
2149        let actual = GitVersion::from_version_stdout(input).unwrap();
2150        let expected = GitVersion {
2151            major: 2,
2152            minor: 18,
2153            patch: 0,
2154        };
2155        assert_eq!(actual, expected);
2156    }
2157
2158    #[test]
2159    fn git_version_version_unreleased() {
2160        let input = "git version 2.55.GIT";
2161        let actual = GitVersion::from_version_stdout(input).unwrap();
2162        let expected = GitVersion {
2163            major: 2,
2164            minor: 55,
2165            patch: 0,
2166        };
2167        assert_eq!(actual, expected);
2168    }
2169}